U. S. Department of Justice
Office of Legal Counsel
Washington, D.C. 20530 June 28, 1995


From: Walter Dellinger
Assistant Attorney General

Re: Adarand

This memorandum sets forth preliminary legal guidance on the implications of the Supreme Court's recent decision in Adarand Constructors, Inc. v. Peņa, 63 U.S.L.W. 4523 (U.S. June 12, 1995), which held that federal affirmative action programs that use racial and ethnic criteria as a basis for decisionmaking are subject to strict judicial scrutiny. The memorandum is not intended to serve as a definitive statement of what Adarand means for any particular affirmative action program. Nor does it consider the prudential and policy questions relevant to responding to Adarand. Rather, it is intended to provide a general overview of the Court's decision and the new standard for assessing the constitutionality of federal affirmative action programs.

Our conclusions can be briefly summarized. Adarand made applicable to federal affirmative action programs the same standard of review, strict scrutiny, that City of Richmond v. J.A. Croson Co., 488 U.S. 469 (1989), applied to state and local affirmative action measures -- with the important caveat that, in this area, Congress may be entitled to greater deference than state and local governments. Although Adarand itself involved contracting, its holding is not confined to that context; rather, it is clear that strict scrutiny will now be applied by the courts in reviewing the federal government's use of race-based criteria in health, education, hiring, and other programs as well.

The Supreme Court in Adarand was careful to dispel any suggestion that it was implicitly holding unconstitutional all federal affirmative action measures employing racial or ethnic classifications. A majority of the Justices rejected the proposit ion that "strict scrutiny" of affirmative action measures means "strict in theory, fatal in fact," and agreed that "the unhappy persistence of both the practice and the lingering effects of racial discrimination against minority groups in this country" ma y justify the use of race-based remedial measures in certain circumstances. 63 U.S.L.W. at 4533. See id. at 4542 (Souter, J., dissenting); id. at 4543 (Ginsburg, J., dissenting). Only two Justices advocated positions that approach a complete ban on aff irmative action.

The Court's decision leaves many questions open -- including the constitutionality of the very program at issue in the case. The Court did not discuss in detail the two requirements of strict scrutiny: the governmental interest underlying an affirmativ e action measure must be "compelling" and the measure must be "narrowly tailored" to serve that interest. As a consequence, our analysis of Adarand's effects on federal action must be based on Croson and the lower court decisions applying s trict scrutiny to state and local programs. It is unclear, however, what differences will emerge in the application of strict scrutiny to affirmative action by the national government; in particular, the Court expressly left open the question of what def erence the judiciary should give to determinations by Congress that affirmative action is necessary to remedy discrimination against racial and ethnic minority groups. Unlike state and local governments, Congress may be able to rely on national findings of discrimination to justify remedial racial and ethnic classifications; it may not have to base such measures on evidence of discrimination in every geographic locale or sector of the economy that is affected. On the other hand, as with state and local governments under Croson, Congress may not predicate race-based remedial measures on generalized, historical societal discrimination.

Two additional questions merit mention at the outset. First, the Court has not resolved whether a governmental institution must have sufficient evidence of discrimination to establish a compelling interest in engaging in race-based remedial action befor e it takes such action. A number of courts of appeals have considered this question in reviewing state and local affirmative action plans after Croson, and all have concluded that governments may rely on "post-enactment" evidence -- that is, evid ence that the government did not consider when adopting the measure, but that reflects evidence of discrimination providing support for the government's determination that remedial action was warranted at the time of adoption. Those courts have said that the government must have had some evidence of discrimination when instituting an affirmative action measure, but that it need not marshal all the supporting evidence at that time. Second, while Adarand makes clear that remedying past discriminati on will in some circumstances constitute a compelling interest sufficient to justify race-based measures, the Court did not address the constitutionality of programs aimed at advancing nonremedial objectives -- such as promoting diversity and inclusion. For example, under Justice Powell's controlling opinion in Regents of the University of California v. Bakke, 438 U.S. 265 (1978), increasing the racial and ethnic diversity of the student body at a university constitutes a compelling interest, beca use it enriches the academic experience on campus. Under strict scrutiny, it is uncertain whether and in what settings diversity is a permissible goal of affirmative action beyond the higher education context. To the extent that affirmative action is us ed to foster racial and ethnic diversity, the government must seek some further objective beyond the achievement of diversity itself.

Our discussion in this memorandum proceeds in four steps. In Section I, we analyze the facts and holding of Adarand itself, the scope of what the Court did decide, and the questions it left unanswered. Section II addresses the strict scrutiny st andards as applied to state and local programs in Croson and subsequent lower court decisions; we consider the details of both the compelling interest and the narrow tailoring requirements Croson mandated. In Section III, we turn to the dif ficult question of how precisely the Croson standards should apply to federal programs, with a focus on the degree of deference courts may give to congressional determinations that affirmative action is warranted. Finally, in an appendix, we sketc h out a series of questions that should be considered in analyzing the validity under Adarand of federal affirmative action programs that employ race or ethnicity as a criterion. The appendix is intended to guide agencies as they begin that proces s.

I. The Adarand Case

A. Facts

Adarand involved a constitutional challenge to a Department of Transportation ("DOT") program that compensates persons who receive prime government contracts if they hire subcontractors certified as small businesses controlled by "socially and eco nomically disadvantaged" individuals. The legislation on which the DOT program is based, the Small Business Act, establishes a government-wide goal for participation of such concerns at "not less than 5 percent of the total value of all prime contract an d subcontract awards for each fiscal year." 15 U.S.C. 644(g)(1). The Act further provides that members of designated racial and ethnic minority groups are presumed to be socially disadvantaged. Id. § 637(a)(5), § 637(d)(2),(3); 13 C.F.R. § 124.105(b) (1).(1) The presumption is rebuttable. 13 C.F.R. §§ 124.111(c)-(d), 124.601-124.609. (2)

In Adarand, a nonminority firm submitted the low bid on a DOT subcontract. However, the prime contractor awarded the subcontract to a minority-owned firm that was presumed to be socially disadvantaged; thus, the prime contractor received additiona l compensation from DOT. 63 U.S.L.W. at 4525. The nonminority firm sued DOT, arguing that it was denied the subcontract because of a racial classification, in violation of the equal protection component of the Fifth Amendment's Due Process Clause. The district court granted summary judgment for DOT. The Court of Appeals for the Tenth Circuit affirmed, holding that DOT's race-based action satisfied the requirements of "intermediate scrutiny," which it determined was the applicable standard of review un der the Supreme Court's rulings in Metro Broadcasting, Inc. v. FCC, 497 U.S. 547 (1990), and Fullilove v. Klutznick, 448 U.S. 448 (1980). See Adarand, 63 U.S.L.W. at 4525.

B. The Holding

By a five-four vote, in an opinion written by Justice O'Connor, the Supreme Court held in Adarand that strict scrutiny is now the standard of constitutional review for federal affirmative action programs that use racial or ethnic classifications a s the basis for decisionmaking. The Court made clear that this standard applies to programs that are mandated by Congress, as well as those undertaken by government agencies on their own accord. 63 U.S.L.W. at 4530. The Court overruled Metro Broadca sting to the extent that it had prescribed a more lenient standard of review for federal affirmative action measures. Id.(3)

Under strict scrutiny, a racial or ethnic classification must serve a "compelling interest " and must be "narrowly tailored" to serve that interest. Id.(4) This is the same standard of review that, under the Supreme Court's decision in City of Richmond v. J.A. Croson Co., 488 U.S. 469 (1989), applies to af firmative action measures adopted by state and local governments. It is also the same standard of review that applies to government classifications that facially discriminate against minorities. 63 U.S.L.W. at 4529, 4531.

In a portion of her opinion joined by Chief Justice Rehnquist, Justice Kennedy, and Justice Thomas, Justice O'Connor sought to "dispel the notion that strict scrutiny is `strict in theory, but fatal in fact'" when it comes to affirmative action. Id. at 4533 (quoting Fullilove, 448 U.S. at 519 (Marshall, J., concurring in the judgment)). While that familiar maxim doubtless remains true with respect to classifications that, on their face, single out racial and ethnic minorities for invidious treat ment,(5) Justice O'Connor's opinion declared that the federal government may have a compelling interest to act on the basis of race to overcome the "persistence of both the practice and lingering effects of racial discriminat ion against minority groups in this country." Id. In this respect, Justice O'Connor's opinion in Adarand tracks her majority opinion in Croson. There, too, the Court declined to interpret the Constitution as imposing a flat ban on affirma tive action by state and local governments. 488 U.S. at 509-11.

Two members of the Adarand majority, Justices Scalia and Thomas, wrote separate concurring opinions in which they took a more stringent position. Consistent with his concurring opinion in Croson, Justice Scalia would have adopted a near-ab solute constitutional bar to affirmative action. Taking issue with Justice O'Connor's proposition that racial classifications may be employed in certain circumstances to remedy discrimination against minorities, Justice Scalia stated that the "government can never have a `compelling interest' in discriminating on the basis of race to `make-up' for past racial discrimination in the opposite direction." 63 U.S.L.W. at 4534 (Scalia, J., concurring in part and concurring in the judgment). (6) According to Justice Scalia, "[i]ndividuals who have been wronged by unlawful racial discrimination should be made whole; but under our Constitution there can be no such thing as either a creditor or a debtor race. That concept is alien to the Constitution's focus on the individual . . . ." Id. The compensation of victims of specific instances of discrimination through "make-whole" relief, which Justice Scalia accepts as legitimate, is not affirmative action, as that term is generally understood. Affirmative action is a group-based remedy: where a group has been subject to discrimination, individual members of the group can benefit from the remedy, even if they have not proved that they have been discriminated against personally. (7) Justice O'Connor's treatment of affirmative action in Adarand is consistent with this understanding.

Although Justice Thomas joined the portion of Justice O'Connor's opinion holding that the government's interest in redressing the effects of discrimination can be sufficiently compelling to warrant the use of remedial racial and ethnic classification s, he apparently agrees with Justice Scalia's rejection of the group-based approach to remedying discrimination. Justice Thomas stated that the "government may not make distinctions on the basis of race," and that it is "irrelevant whether a government's racial classifications are drawn by those who wish to oppress a race or by those who have a sincere desire to help those thought to be disadvantaged." Id. (Thomas, J., concurring in part and concurring in the judgment).

The four dissenting Justices in Adarand (Justices Stevens, Souter, Ginsburg, and Breyer) (8) would have reaffirmed the intermediate scrutiny standard of review for congressionally authorized affirmative action measures esta blished in Metro Broadcasting, and would have sustained the DOT program on the basis of Fullilove, where the Court upheld federal legislation requiring grantees to use at least ten percent of certain grants for public works projects to procu re goods and services from minority businesses. Justices Stevens and Souter argued that the DOT program was more narrowly tailored than the legislation upheld in Fullilove. 63 U.S.L.W. at 4539-41 (Stevens, J., dissenting); id. at 4542 (Souter. J. , dissenting). All four dissenters stressed that there is a constitutional distinction between racial and ethnic classifications that are designed to aid minorities and classifications that discriminate against them. As Justice Stevens put it, there is a difference between a "No Trespassing" sign and a "welcome mat." Id. at 4535 (Stevens, J., dissenting). See id. ("an attempt by the majority to exclude members of a minority race from a regulated market is fundamentally different from a [race-based] su bsidy that enables a relatively small group of [minorities] to enter that market."); see also id. at 4543 (Souter, J., dissenting); id. at 4544 (Ginsburg, J., dissenting). For the dissenters, Justice O'Connor's declaration that strict scrutiny of affirma tive action programs is not "fatal in fact" signified a "common understanding" among a majority of the Court that those differences do exist, and that affirmative action may be entirely proper in some cases. Id. at 4543 (Ginsburg, J., dissenting). In Ju stice Ginsburg's words, the "divisions" among the Justices in Adarand "should not obscure the Court's recognition of the persistence of racial inequality and a majority's acknowledgment of Congress' authority to act affirmatively, not only to end d iscrimination, but also to counteract discrimination's lingering effects." Id. The dissenters also emphasized that there is a "significant difference between a decision by the Congress of the United States to adopt an affirmative-action program and such a decision by a State or a municipality." Id. at 4537 (Stevens, J., dissenting); id. at 4542 (Souter, J., dissenting). They stressed that unlike state and local governments, Congress enjoys express constitutional power to remedy discrimination against minorities; therefore, it has more latitude to engage in affirmative action than do state and local governments. Id. at 4538 (Stevens, J., dissenting). Justice Souter noted that the majority opinion did not necessarily imply a contrary view. Id. at 454 2 (Souter, J., dissenting).

Thus, there were at most two votes in Adarand (Justices Scalia and Thomas) for anything that approaches a blanket prohibition on race-conscious affirmative action. Seven justices confirmed that federal affirmative action programs that use race or ethnicity as a decisional factor can be legally sustained under certain circumstances.

C. Scope of Adarand

Although Adarand involved government contracting, it is clear from the Supreme Court's decision that the strict scrutiny standard of review applies whenever the federal government voluntarily adopts a racial or ethnic classification as a basis for decisionmaking. (9) Thus, the impact of the decision is not confined to contracting, but will reach race-based affirmative action in health and education programs, and in federal employment. (10) Furthermore, Adarand was not a "quota" case: its standards will apply to any classification that makes race or ethnicity a basis for decisionmaking. (11) Mere outreach and recruitment efforts, however, typicall y should not be subject to the Adarand standards. Indeed, post-Croson cases indicate that such efforts are considered race-neutral means of increasing minority opportunity. (12) In some sense, of course, the targeti ng of minorities through outreach and recruitment campaigns involves race-conscious action. But the objective there is to expand the pool of applicants or bidders to include minorities, not to use race or ethnicity in the actual decision. If the governm ent does not use racial or ethnic classifications in selecting persons from the expanded pool, Adarand ordinarily would be inapplicable. (13)

Adarand does not require strict scrutiny review for programs be nefitting Native Americans as members of federally recognized Indian tribes. In Morton v. Mancari, 417 U.S. 535 (1974), the Supreme Court applied rational basis review to a hiring preference in the Bureau of Indian Affairs for members of federally recognized Indian tribes. The Court reasoned that a tribal classification is "political rather than racial in nature," because it is "granted to Indians not as a discrete racial group, but, rather, as members of quasi-sovereign tribal entities." Id. at 554. See id. at 553 n.24.

Adarand did not address the appropriate constitutional standard of review for affirmative action programs that use gender classifications as a basis for decisionmaking. Indeed, the Supreme Court has never resolved the matter. (14) However, both before and after Croson, nearly all circuit court decisions have applied intermediate scrutiny to affirmative action measures that benefit women. (15) The Sixth Circuit is the only court that has equated racial and gender classifications: purporting to rely on Croson, it held that gender-based affirmative action measures are subject to strict scrutiny. (16) That holding has been criticized by other cour ts of appeals, which have correctly pointed out that Croson does not speak to the appropriate standard of review for such measures. (17)

D. Open Questions on Remand

Adarand did not determine the constitutionality of any particular federal affirmative action program. In fact, the Supreme Court did not determine the validity of the federal legislation, regulations, or program at issue in Adarand itself. Instead, the Court remanded the case to the Tenth Circuit for a determination of whether the measures satisfy strict scrutiny.

Adarand left open the possibility that, even under strict scrutiny, programs statutorily prescribed by Congress may be entitled to greater deference than programs adopted by state and local governments. This is a theme that some of the Justices h ad explored in prior cases. For example, in a portion of her Croson opinion joined by Chief Justice Rehnquist and Justice White, Justice O'Connor wrote that Congress may have more latitude than state and local governments in utilizing affirmative action. And in his concurrence in Fullilove, Justice Powell, applying strict scrutiny, upheld a congressionally mandated program, and in so doing, said that he was mindful that Congress possesses broad powers to remedy discrimination nationwide. In any event, in Adarand, the Court said that it did not have to resolve whether and to what extent courts should pay special deference to Congress in evaluating federal affirmative action programs under strict scrutiny.

Aside from articulating the components of the strict scrutiny standard, the Court's decision inAdarand provides little explanation of how the standard should be applied. For more guidance, one needs to look to Croson and lower court decis ions applying it. That exercise is important because Adarand basically extends the Croson rules of affirmative action to the federal level -- with the caveat that application of those rules might be somewhat less stringent where affirmative action is undertaken pursuant to congressional mandate.

II. The Croson Standards

In Croson, the Supreme Court considered a constitutional challenge to a Richmond, Virginia ordinance that required prime contractors who received city contracts to subcontract at least thirty percent of the dollar amount of those contracts to busi nesses owned and controlled by members of specified racial and ethnic minority groups -- commonly known as minority business enterprises ("MBEs"). The asserted purpose of Richmond's ordinance was to remedy discrimination against minorities in the local c onstruction industry.

Croson marked the first time that a majority of the Supreme Court held that race-based affirmative action measures are subject to strict scrutiny. (18) Justice O'Connor's opinion in Croson(19) said that "the purpose of strict scrutiny is to `smoke out' illegitimate uses of race by assuring that the legislative body is pursuing a goal important enough to warrant use of a highly suspect tool. The test also ensures that the means chosen `fit' this compelling goal so closely that there is little or no possibility that the motive for the classification was illegitimate racial prejudice or stereotype." 488 U.S. at 493 (plurality opinion). See also id. at 520 (Scalia, J., concurring in the judgment) ("[S]trict scrutiny must be applied to all governmental classifications by race, whether or not its asserted purpose is `remedial' or `benign.'"). In short, the compelling interest inquiry centers on "ends" and asks why the government i s classifying individuals on the basis of race or ethnicity; the narrow tailoring inquiry focuses on "means" and asks how the government is seeking to meet the objective of the racial or ethnic classification.

Applying strict scrutiny, the Court held that (a) the Richmond MBE program did not serve a "compelling interest" because it was predicated on insufficient evidence of discrimination in the local construction industry, and (b) it was not "narrowly tailore d" to the achievement of the city's remedial objective.

A. Compelling Governmental Interest

1. Remedial Objectives

Justice O'Connor's opinion in Croson stated that remedying the identified effects of past discrimination may constitute a compelling interest that can support the use by a governmental institution of a racial or ethnic classification. This discri mination could fall into two categories. First, the government can seek to remedy the effects of its own discrimination. Second, the government can seek to remedy the effects of discrimination committed by private actors within its jurisdiction, where t he government becomes a "passive participant" in that conduct, and thus helps to perpetuate a system of exclusion. 488 U.S. at 492 (plurality opinion); id. at 519 (Kennedy, J., concurring in part and concurring in the judgment). In either category, the remedy may be aimed at ongoing patterns and practices of exclusion, or at the lingering effects of prior discriminatory conduct that has ceased. See Adarand, 63 U.S.L.W. at 4542 (Souter, J., dissenting) ("The Court has long accepted the view that constitutional authority to remedy past discrimination is not limited to the power to forbid its continuation, but extends to eliminating those effects that would otherwise persist and skew the operation of public systems even in the absence of current in tent to practice any discrimination.").

Croson requires the government to identify with precision the discrimination to be remedied. The fact and legacy of general, historical societal discrimination is an insufficient predicate for affirmative action: "While there is no doubt that the sorry history of both private and public discrimination in this country has contributed to a lack of opportunities for black entrepreneurs, this observation, standing alone, cannot justify a rigid racial quota in the awarding of public contracts in Richm ond, Virginia." 488 U.S. at 499. See id. at 505 ("To accept Richmond's claim that past societal discrimination alone can serve as the basis for rigid racial preferences would be to open the door to competing claims for `remedial relief' for every disadv antaged group."). Similarly, "amorphous" claims of discrimination in certain sectors and industries are inadequate. Id. at 499 ("[A]n amorphous claim that there has been past discrimination in a particular industry cannot justify the use of an unyieldin g racial quota."). Such claims "provide[] no guidance for [the government] to determine the precise scope of the injury it seeks to remedy, and would have "no logical stopping point." Id. at 498 (internal quotations omitted). The Court indicated that i ts requirement that the government identify with specificity the effects of past discrimination anchors remedial affirmative action measures in the present. It declared that "[i]n the absence of particularized findings" of discrimination, racial and ethn ic classifications could be "ageless in their reach into the past, and timeless in their ability to affect the future." Id. at 498. (internal quotations omitted).

The Court in Croson did not require a judicial determination of discrimination in order for a state or local government to adopt remedial racial or ethnic classifications. Rather, relying on Justice Powell's plurality opinion in Wygant v. Jack son Board of Education, 476 U.S. 267 (1986), the Court said that the government must have a "`strong basis in evidence for its conclusion that remedial action was necessary.'" Croson, 488 U.S. at 500 (quoting Wygant, 476 U.S. at 277). The Court then suggested that this evidence should approach "a prima facie case of a constitutional or statutory violation" of the rights of minorities. 488 U.S. at 500. (20) Notably, the Court said that significant statis tical disparities between the level of minority participation in a particular field and the percentage of qualified minorities in the applicable pool could permit an inference of discrimination that would support the use of racial and ethnic classificatio ns intended to correct those disparities. Id. at 507. See id. at 501 ("There is no doubt that where gross statistical disparities can be shown, they alone in a proper case may constitute prima facie proof of a pattern or practice of discrimination.") (i nternal quotations omitted). But the Court said that a mere underrepresentation of minorities in a particular sector or industry when compared to general population statistics is an insufficient predicate for affirmative action. Id. ("When special quali fications are required to fill particular jobs, comparisons to the general population (rather than to the smaller group of individuals who may possess the necessary qualifications) may have little probative value.") (internal quotations omitted).

Applying its "strong basis in evidence" test, the Court held that the statistics on which Richmond based its MBE program were not probative of discrimination in contracting by the city or local contractors, but at best reflected evidence of general societ al discrimination. Richmond had relied on limited testimonial evidence of discrimination, supplemented by statistical evidence regarding: (i) the disparity between the number of prime contracts awarded by the city to minorities during the years 1978-83 (less than one percent) and the city's minority population (fifty percent), and (ii) the extremely low number of MBEs that were members of local contractors' trade associations. The Court found that this evidence was insufficient. It said that more prob ative evidence would have compared, on the one hand, the number of qualified MBEs in the local labor market with, on the other hand, the number of city contracts awarded to MBEs and the number of MBEs in the local contractors' associations.

In Adarand, Justice O'Connor's opinion noted that "racial discrimination against minority groups in this country is an unfortunate reality," and as an example, it pointed to the "pervasive, systematic, and obstinate discriminatory conduct" that und erpinned the court-ordered affirmative action measures that were upheld in United States v. Paradise, 480 U.S. 149 (1987). 63 U.S.L.W. at 4533 (internal quotations omitted). (21) Her opinion did not say, however, th at only overwhelming evidence of the sort at issue in Paradise can justify affirmative action. Again, Croson indicates that what is required is a "strong basis in evidence" to support the government's conclusion that race-based remedial act ion is warranted, and that such evidence need only approach a prima facie showing of discrimination against minorities. 488 U.S. at 500. The factual predicate in Paradise plainly exceeded a prima facie showing. Post-Croson lower court dec isions support the conclusion that the requisite factual predicate for race-based remedial action does not have to rise to the level of discrimination in Paradise.

The Court in Croson left open the question whether a government may introduce statistical evidence showing that the pool of qualified minorities would have been larger "but for" the discrimination that is to be remedied. Post-Croson lower c ourt decisions have indicated that such evidence can be probative of discrimination. (22)

Croson also did not discuss the weight to be given to anecdotal evidence of discrimination that a government gathers thro ugh complaints filed with it by minorities or through testimony in public hearings. Richmond had relied on such evidence as additional support for its MBE plan, but the Court discounted it. Post-Croson lower court cases, however, have said that a necdotal evidence can buttress statistical proof of discrimination. (23)

In addition, Croson did not discuss which party has the ultimate burden of persuasion as to the constitutionality of an affirmative action program when it is challenged in court. Prior to Croson, the Supreme Court had spelled out the following evidentiary rule: while the entity defending a remedial affirmative action measure bears the initial burden of production to show that the m easures are supported by "a strong basis in evidence," the "ultimate burden" of proof rests upon those challenging the measure to demonstrate that it is unconstitutional. Wygant, 476 U.S. at 277-78 (plurality opinion). ( 24) Lower courts consistently have said that nothing in Croson disturbs this evidentiary rule. (25)

Finally, and perhaps most significantly, Croson did not resolve whether a government must have suffi cient evidence of discrimination at hand before it adopts a racial classification, or whether "post-hoc" evidence of discrimination may be used to justify the classification at a later date -- for example, when it is challenged in litigation. The Court d id say that governments must "identify [past] discrimination with some specificity before they may use race-conscious relief." 488 U.S. at 504. However, every court of appeals to consider the question has allowed governments to use "post-enactment" evid ence to justify affirmative action -- that is, evidence that the government did not consider when adopting a race-based remedial measure, but that nevertheless reflects evidence of discrimination providing support for the determination that remedial actio n was warranted at the time of adoption. (26) Those courts have interpreted Croson as requiring that a government have some evidence of discrimination prior to embarking on remedial race-conscious action, but not that it marshal all such evidence at that time. (27)

2. Nonremedial Objectives

Because Richmond defended its MBE program on remedial grounds, the Court in Croson did not explicitly address if and when affirmative action may be adopted for "nonremedial" objectives, such as promoting racial diversity and inclusion. The same is true of the majority opinion in Adarand, since the program at issue in that case also is said to be remedial. In h is Adarand dissent, Justice Stevens said that the majority's silence on the question does not foreclose the use of affirmative action to serve nonremedial ends. 63 U.S.L.W. at 4539 (Stevens, J., dissenting). Thus, in the wake of Croson and Adarand, there are substantial questions as to whether and in what settings nonremedial objectives can constitute a compelling interest. (28)

To date, there has never been a majority opinion for the Supreme Cou rt that addresses the question. The closest the Court has come in that regard is Justice Powell's separate opinion in Regents of the University of California v. Bakke, 438 U.S. 265 (1978), which said that a university has a compelling interest in taking the race of applicants into account in its admissions process in order to foster greater diversity among the student body. (29) According to Justice Powell, this would bring a wider range of perspectives to the campu s, and in turn, would contribute to a more robust exchange of ideas -- which Justice Powell said was the central mission of higher education and in keeping with the time-honored First Amendment value in academic freedom. See id. at 311-14. (30) Since Bakke, Justice Stevens has been the most forceful advocate on the Court for nonremedial affirmative action measures. He has consistently argued that affirmative action makes just as much sense when it promotes an inter est in creating a more inclusive and diverse society for today and the future, as when it serves an interest in remedying past wrongs. See Adarand, 63 U.S.L.W. at 4539 (Stevens, J., dissenting); Croson, 488 U.S. at 511-12 & n.1 (Stevens, J. , concurring); Johnson v. Transportation Agency, 480 U.S. 616, 646-47 (1987) (Stevens, J., concurring); Wygant, 476 U.S. at 313-15 (Stevens, J., dissenting). As a circuit judge in a case involving an ostensibly remedial affirmative action measure, Justice Ginsburg announced her agreement with Justice Stevens' position "that remedy for past wrong is not the exclusive basis upon which racial classifications may be justified." O'Donnell Constr. Co. v. District of Columbia, 963 F.2d 42 0, 429 (D.C. Cir. 1992) (Ginsburg, J., concurring) (citing Justice Stevens' concurrence in Croson, 488 U.S. at 511).

In Metro Broadcasting, the majority relied on Bakke and Justice Stevens' vision of affirmative action to uphold FCC affirmative action programs in the licensing of broadcasters on nonremedial grounds; the Court said that diversification of ownership of broadcast licenses was a permissible objective of affirmative action because it serves the larger goal of exposing the nation to a greater diversity of perspectives over the nation's radio and television airwaves. 497 U.S. at 567-68. The Co urt reached that conclusion under intermediate scrutiny, however, and thus did not hold that the governmental interest in seeking diversity in broadcasting is "compelling." Adarand did not overrule the result in Metro Broadcasting -- a poin t not lost on Justice Stevens. See Adarand, 63 U.S.L.W. at 4539 (Stevens, J., dissenting) ("The majority today overrules Metro Broadcasting only insofar as it" is inconsistent with the holding that federal affirmative action measures are su bject to strict scrutiny. "The proposition that fostering diversity may provide a sufficient interest to justify [a racial or ethnic classification] is not inconsistent with the Court's holding today -- indeed, the question is not remotely presented in th is case . . . .").

On the other hand, portions of Justice O'Connor's opinion in Croson and her dissenting opinion in Metro Broadcasting appear to cast doubt on the validity of nonremedial affirmative action programs. In one passage in her opinion in Croson , Justice O'Connor stated that affirmative action must be "strictly reserved for the remedial setting." Id. at 493 (plurality opinion). Echoing that theme in her dissenting opinion (joined by Chief Justice Rehnquist and Justices Kennedy and Scalia) in Metro Broadcasting, Justice O'Connor urged the adoption of strict scrutiny for federal affirmative action measures, and asserted that under that standard, only one interest has been "recognized" as compelling enough to justify racial classificat ions: "remedying the effects of racial discrimination." 497 U.S. at 612. Justice Kennedy's separate dissent in Metro Broadcasting was also quite dismissive of non-remedial justifications for affirmative action; he criticized the majority opinion for "allow[ing] the use of racial classifications by Congress untied to any goal of addressing the effects of past race discrimination"). Id. at 632 (Kennedy, J., dissenting).

Nowhere in her Croson and Metro Broadcasting opinions did Justice O'Connor expressly disavow Justice Powell's opinion in Bakke. Accordingly, lower courts have assumed that Justice O'Connor did not intend to discard Bakke.(31) That proposition is supported by Justice O'Connor's own concurring opinion in Wygant v. Jackson Board of Education, 476 U.S. 267 (1986), in which she expressed approval of Justice Powell's view that fostering racial and ethnic diversity in higher education is a compelling interest. Id. at 286. Furthermore, in Wygant, Justice O'Connor said that there might be governmental interests other than remedying discrimination and promoting diversity in higher educatio n that might be sufficiently compelling to support affirmative action. Id. For example, Justice O'Connor left open the possibility that promoting racial diversity among the faculty at primary and secondary schools could count as a compelling interest. Id. at 288 n*. In his Wygant dissent, Justice Stevens argued that this is a permissible basis for affirmative action. Id. at 313-15 (Stevens, J., dissenting).

On the assumption that Bakke remains the law, it is clear that to the extent affirmative action is used to foster racial and ethnic diversity, the government must seek some further objective, beyond the mere achievement of diversity itself. (32) As Bakke teaches, in higher education, that asserted goal is the enrichment of the academic experience. And according to the majority in Metro Broadcasting, the asserted independent goal that justifies diversif ying the owners of broadcast licenses is adding variety to the perspectives that are communicated in radio and television. That same kind of analysis must be applied to efforts to promote racial and ethnic diversity in other settings.

For instance, diversification of the ranks in a law enforcement agency arguably serves vital public safety and operational needs, and thus enhances the agency's ability to carry out its functions effectively. See Wygant, 476 U.S. at 314 (Stevens, J., dissenting) ("[I]n law enforcement . . . in a city with a recent history of racial unrest, the superintendent of police might reasonably conclude that an integrated police force could develop a better relationship with the community and thereby do a m ore effective job of maintaining law and order than a force composed only of whites."); Paradise, 480 U.S. at 167 n.18 (plurality opinion) (noting argument that race-conscious hiring can "restore[] community trust in the fairness of law enforcement and facilitate[] effective police service by encouraging citizen cooperation").(33) It is more difficult to identify any independent goal that may be attained by diversifying the racial mix of public contractors. Justice Stevens concurred in the judgment in Croson on precisely that ground. Citing his own Wygant dissent, Justice Stevens contrasted the "educational benefits to the entire student body" that he said could be achieved through faculty diversity w ith the minimal societal benefits (other than remedying past discrimination, a predicate that he said was not supported by the evidence in Croson) that would flow from a diversification of the contractors with whom a municipality does business. Se e Croson, 488 U.S. at 512-13 (Stevens, J., concurring in part and concurring in the judgment). Furthermore, the Court has stated that the desire to develop a growing class of successful minority entrepreneurs to serve as "role models" in the minority com munity is not, on its own, a valid basis for a racial and ethnic classification. See Croson, 488 U.S. at 497 (citing Wygant, 476 U.S. at 276 (plurality opinion)); see also Wygant, 476 U.S. at 288 n* (O'Connor, J., concurring).

Diversification of the health services profession was one of the stated predicates of the racial and ethnic classifications in the medical school admissions program at issue in Bakke. The asserted independent goal was "improving the delivery of he alth-care services to communities currently underserved." Bakke, 438 U.S. at 310. Justice Powell said that "[i]t may be assumed that in some situations a State's interest in facilitating the health care of its citizens is sufficiently compelling to support the use of a suspect classification." Id. The problem in Bakke, however, was that there was "virtually no evidence" that the preference for minority applicants was "either needed or geared to promote that goal." Id. (34)

Assuming that some nonremedial objectives remain a legitimate basis for affirmative action after Adarand, there is a question of the nature of the showing that may be necessary to support racial and ethnic classifications that ar e premised on such objectives. In higher education, the link between the diversity of the student body and the diversity of viewpoints on the campus does not readily lend itself to empirical proof. Justice Powell did not require any such evidence in Bakke. He said that the strong First Amendment protection of academic freedom that allows "a university to make its own judgments as to education includes the selection of its student body." Bakke, 438 U.S. at 312. A university is thus due s ome discretion to conclude that a student "with a particular background -- whether it be ethnic, geographic, culturally advantaged or disadvantaged -- may bring to a professional school of medicine experiences, outlooks, and ideas that enrich the training of its student body and better equip its graduates to render with understanding their vital service to humanity." Id. at 314.

It could be said that this thesis is rooted in a racial stereotype, one that presumes that members of racial and ethnic minority groups have a "minority perspective" to convey. As Justice O'Connor stated in Croson, a driving force behind strict s crutiny is to ensure that racial and ethnic classifications are not motivated by "stereotype." Croson, 488 U.S. at 493 (plurality opinion). There are sound arguments to support the contention that seeking diversity in higher education rests on val id assumptions. The thesis does not presume that all individuals of a particular race or ethnic background think and act alike. Rather, it is premised on what seems to be a common sense proposition that in the aggregate, increasing the diversity of the student body is bound to make a difference in the array of perspectives communicated at a university. See Metro Broadcasting, 497 U.S. at 579 ("The predictive judgment about the overall result of minority entry into broadcasting is not a rigid assumption about how minority owners will behave in every case but rather is akin to Justice Powell's conclusion in Bakke that greater admission of minorities would contribute, on average, to the robust exchange of ideas.") (internal quotations omitted). No netheless, after Croson and Adarand, a court might demand some proof of a nexus between the diversification of the student body and the diversity of viewpoints expressed on the campus. (35) Likewise, a court m ay demand a factual predicate to support the proposition that greater diversity in a law enforcement agency will serve the operational needs of the agency and improve its performance, (36) or that minority health care profes sionals are more likely to work in medically underserved communities. (37)

B. Narrow Tailoring Test

In addition to advancing a compelling goal, any governmental use of race must also be "narrowly tailored." There appear to be two underlying purposes of the narrow tailoring test: first, to ensure that race-based affirmative action is the product of careful deliberation, not hasty decisionmaking; and, second, to ensure that such action is truly necessary, and that less intrusive, efficacious means to the end are unavailable. As it has been applied by the courts, the factors that typically make up the "narrow tailoring" test are as follows: (i) whether the government considered race-neutral alternatives before resorting to race-conscious action; (ii) the scope of the affirmative action program, and whether there is a waiver mechanism that facilitates the narrowing of the program's scope; (iii) the manner in which is used, that is, whether race is a factor in d etermining eligibility for a program or whether race is just one factor in the decisionmaking process; (iv) the comparison of any numerical target to the number of qualified minorities in the relevant sector or industry; (v) the duration of the program an d whether it is subject to periodic review; and (vi) the degree and type of burden caused by the program. In Adarand, the Supreme Court referred to its previous affirmative action decisions for guidance on what the narrow tailoring test entails. It specifically mentioned that when the Tenth Circuit reviewed the DOT program at issue in Adarand under intermediate scrutiny, it had not addressed race-neutral alternatives or the duration of the program.

Before describing each of the components, three general points about the narrow tailoring test deserve mention. First, it is probably not the case that an affirmative action measure has to satisfy every factor. A strong showing with respect to most of t he factors may compensate for a weaker showing with respect to others.

Second, all of the factors are not relevant in every case. For example, the objective of the program may determine the applicability or weight to be given a factor. The factors may play out differently where a program is nonremedial.

Third, the narrow tailoring test should not necessarily be viewed in isolation from the compelling interest test. To be sure, the inquiries are distinct: as indicated above, the compelling interest inquiry focuses on the ends of an affirmative action me asure, whereas the narrow tailoring inquiry focuses on the means. However, as a practical matter, there may be an interplay between the two. There is some hint of this in Croson. In several places, the Court said that the weak predicate of discr imination on which Richmond acted could not justify the adoption of a rigid racial quota -- which suggests that if Richmond had opted for some more flexible measure the Court might have been less demanding when reviewing the evidence of discrimination. B y the same token, the more compelling the interest, perhaps less narrow tailoring is required. For example, in Sheet Metal Workers v. EEOC, 478 U.S. 421 (1986), and United States v. Paradise, 480 U.S. 149 (1987), the Supreme Court upheld wh at on their face appear to be rather rigid classifications to remedy egregious and persistent discrimination.

However, it bears emphasizing that the Supreme Court has never explicitly recognized any trade-off between the compelling interest and narrow tailoring tests. It is also far from clear that the Court in Croson would have found that a more flexible MBE program, supported by the generalized evidence of discrimination on which Richmond relied, could withstand strict scrutiny. In addition, the membership of the Court has changed dramatically in the years since Sheet Metal Workers and Paradi se. Both cases were decided by five-four margins, and only one member of the majority (Justice Stevens) remains. And while Justice O'Connor agreed with the majority in Sheet Metal Workers and Paradise that ample evidence of deeply entr enched discrimination gave rise to a very weighty interest in race-based action, she dissented on the ground that the particular remedies selected were too rigid.

1. Race-Neutral Alternatives

In Croson, the Supreme Court said that the Richmond MBE program was not "narrowly tailored," in part because the city apparently had not considered race-neutral means to increase minority participation in contracting before adopting its race-based measure. The Court reasoned that because minority businesses tend to be smaller and less-established, providing race-neutral financial and technical assistance to small and/or new firms and relaxing bonding requirements might achieve the desired remedial results in public contracting -- increasing opportunities for minority businesses. 488 U.S. at 507, 510. Justice Scalia suggested an even more aggressive idea: "adopt a preference for small businesses, or even for new businesses -- which would make it easier for those previously excluded by discrimination to enter the field. Such programs may well have a racially disproportionate impact, but they are not based on race." Id. at 526 (Scalia, J., concurring). As such, they would not be subjected to str ict scrutiny.

The Court in Croson did not specify the extent to which governments must consider race-neutral measures before resorting to race-conscious action. It would seem that the government need not first exhaust race-neutral alternatives, but only give t hem serious attention. (38) This principle would comport with the purposes of ensuring that race-based remedies are used only when, after careful consideration, a government has concluded that less intrusive means would not work. It also comports with Justice Powell's view that in the remedial setting, the government need not use the "least restrictive means" where they would not accomplish the desired ends as well. See Fullilove, 448 U.S. at 508 (Powell, J., concurring); see also Wygant, 476 U.S. at 280 n.6 (plurality opinion of Justice Powell) (narrow tailoring requirement ensures that "less restrictive means" are used when they would promote the objectives of a racial classification "about as well") (internal quotation s omitted). (39)

This approach gives the government a measure of discretion in determining whether its objectives could be accomplished through some other avenue. In addition, under this approach, the government may not be obliged to consider race-neutral alternatives every time that it adopts a race-conscious measure in a particular field. In some situations, the government may be permitted to draw upon a previous consideration of race-neutral alternatives that it und ertook prior to adopting some earlier race-based measure. (40) In the absence of prior experience, however, a government should consider race-neutral alternatives at the time it adopts a racial or ethnic classification. Mo re fundamentally, even where race-neutral alternatives were considered, a court might second-guess the government if the court believes that an effective race-neutral alternative is readily available and hence should have been tried. See Metro Broadca sting, 497 U.S. at 625 (O'Connor, J., dissenting) (FCC affirmative action programs are not narrowly tailored, in part, because "the FCC has never determined that it has any need to resort to racial classifications to achieve its asserted interest, and it has employed race-conscious means before adopting readily available race-neutral, alternative means"); United States v. Paradise, 480 U.S. at 199-200 (O'Connor, J., dissenting) (district court's race-based remedial order was not narrowly tailor ed because the court "had available several alternatives" that would have achieved the objectives in a less intrusive manner). (41)

2. Scope of Program/Administrative Waivers

Justice O'Connor's opinion for the Court in Croson criticized the scope of Richmond's thirty percent minority subcontracting requirement, calling it a "rigid numerical quota" that did not permit consideration, through some form of administrative waiver me chanism, of whether particular individuals benefiting from the ordinance had suffered from the effects of the discrimination that the city was seeking to remedy. 488 U.S. at 508. At first blush, this criticism of the Richmond plan may appear to conflict with previous Court decisions, joined by Justice O'Connor, that held that race-based remedial measures need not be limited to persons who were the victims of discrimination. (See supra p. 5.) Upon closer reading, however, Croson should not be interpret ed as introducing a "victims-only" requirement through the narrow tailoring test. (42) The Court's rejection in Adarand of Justice Scalia's position that compensation is due only to individuals who have been discrimi nated against personally provides further confirmation that Croson did not impose any such requirement.

The Court's focus in Croson on individualized consideration of persons seeking the benefit of a racial classification appears to have been animated by three separate concerns about the scope of the Richmond plan. First, the Court indicated that in order for a remedial affirmative action program to be narrowly tailored, its beneficiaries must be members of groups that were the victims of discrimination. The Court faulted the Richmond plan because it was intended to remedy discrimination against Af rican-American contractors, but included among its beneficiaries Hispanics, Asian-Americans, Native-Americans, Eskimos, and Aleuts -- groups for which Richmond had proffered "absolutely no evidence of past discrimination." Id. at 506. Therefore, the Cou rt said, even if the Richmond MBE program was "`narrowly tailored' to compensate African-American contractors for past discrimination, one may legitimately ask why they are forced to share this `remedial relief' with an Aleut citizen who moves to Richmond tomorrow?" Id. (43) Second, the Court said that the Richmond plan was not even narrowly tailored to remedy discrimination against black contractors because "a successful black entrepreneur . . . from anywhere in the count ry" could reap its benefits. Id. at 508. That is, the geographic scope of the plan was not sufficiently tailored. (44) Third, the Court contrasted the "rigidity" of the Richmond plan with the flexible waiver mechanism in the ten percent minority participation requirement that was upheld in Fullilove. As the Court in Croson described it, the requirement in Fullilove could be waived where a minority business charged a "higher price [that] was not attributable to the effect s of past discrimination." Id. See Fullilove, 448 U.S. at 488 (plurality opinion). The theory is that where a business is struggling to overcome discrimination, it may not have the capacity to submit a competitive bid. That an effective waiver provis ion allows for "individualized consideration" of a particular minority contractor's bid does not mean that the contractor has to be a "victim" of a specific instance of discrimination. It does mean that if the contractor is wealthy and has entered the ma instream of contractors in the community, a high bid might not be traceable to the discrimination that a racial or ethnic classification is seeking to redress. Instead, such a bid might reflect an effort to exploit the classification. (45)

3. Manner in Which Race is Used

The Court's attack on the "rigidity" of the Richmond ordinance also implicates another common refrain in affirmative action jurisprudence: the manner in which race is used is an integral part of the narrow tailoring requirement. The clearest statement of the Court's somewhat mixed messages in this area is that programs that make race or ethnicity a requirement of eligibility for particular positions or benefits are less likely to survive constitutional cha llenge than programs that merely use race or ethnicity as one factor to be considered under a program open to all races and ethnic groups. (46)

Two types of racial classifications are subject to criticism as being too r igid. First and most obvious is an affirmative action program in which a specific number of positions are set aside for minorities. The prime example is the medical school admissions program that the Court invalidated in Bakke. Justice Powell's pivotal opinion in the case turned squarely on the fact that the program reserved sixteen percent of the slots at the medical school for members of racial and ethnic minority groups. Another example of this type of classification is the program upheld in Fullilove. It provides that, except where the Secretary of Commerce determines otherwise, at least ten percent of the amount of federal grants for certain public works projects must be expended by grantees to purchase goods or services from minority-own ed businesses. 42 U.S.C. 6705(f)(2).

The second type of classification that is vulnerable to attack on flexibility grounds is a program in which race or ethnicity is the sole or primary factor in determining eligibility. One example is the FCC's "distress sale" program, which allows a broadc aster whose qualifications have been called into question to transfer his or her license prior to an FCC revocation hearing, provided the transferee is a minority-owned business. (47) Another example of affirmative action p rograms in which race or ethnicity is a requirement of eligibility are college scholarships that are reserved for minorities. (48)

Under both types of classifications, persons not within the designated categories are ren dered ineligible for certain benefits or positions. (49) Justice Powell's opinion in Bakke rested on the fact that the admissions program at issue was a quota that saved places for minorities solely on the basis of t heir race. (50) As Justice Powell put it, such a program tells applicants who are not Negro, Asian, or Chicano that they are totally excluded from a specific percentage of the seats in an entering class. No matter how str ong their qualifications, quantitative and extracurricular, including their own potential for contribution to educational diversity, they are never afforded the chance to compete with applicants from the preferred groups for the special admissions seats.

438 U.S. at 319. Justice Powell contrasted admissions programs that require decisions based "solely" on race and ethnicity, id. at 315, with programs in which race or ethnic background is simply one factor among many in the admissions decision. Justice Powell said that in the latter type of program, "race or ethnic background may be deemed a `plus' in a particular applicant's file, yet it does not insulate the individual from comparison with all other candidates for the available seats." Id. at 317. I n Justice Powell's view, such programs are sufficiently flexible to meet the narrow tailoring requirement.

This line of reasoning also resonates in Johnson v. Transportation Agency, 480 U.S. 616 (1987). There, the Supreme Court upheld an affirmative action plan under which a state government agency considered the gender of applicants(51) as one factor in making certain promotion decisions. The Court noted that the plan "set aside no positions for women," but simply established goals for female representation that were not "construed" by the agency as "quotas." Id. at 63 8. The Court further observed that the plan "merely authorize[d] that consideration be given to affirmative action concerns when evaluating qualified applicants." Id. The Court stressed that in the promotion decision in question, "sex . . . was but one of numerous factors [that were taken] into account." Id. The agency's plan "thus resemble[d]" the type of admissions program "approvingly noted by Justice Powell" in Bakke: it "requires women to compete with all other qualified applicants. No persons are automatically excluded from consideration; all are able to have their qualifications weighed against those of other applicants." Id. See also id. at 656-57 (O'Connor, J., concurring in judgment) (agency's promotion decision was not made "sol ely on the basis of sex;" rather, "sex was simply used as a `plus factor'").

Finally, Croson itself touches on the point. The Court said that in the absence of a waiver mechanism that permitted individualized consideration of persons seeking a share of city contracts pursuant to the requirement that thirty percent of the d ollar value of prime contracts go to minority subcontractors, the Richmond plan was "problematic from an equal protection standpoint because [it made] the color of an applicant's skin the sole relevant consideration." 488 U.S. at 508.

4. Comparison of Numerical Target to Relevant Market

Where an affirmative action program is justified on remedial grounds, the Court has looked at the size of any numerical goal and its comparison to the relevant labor market or industry. This factor involves choosing the appropriate measure of comparison. In Croson, Richmond defended its thirty percent minority subcontracting requirement on the premise that it was halfway between .067 percent -- the percentage of city contracts awarded to African-Americans during the years 1978-83 -- and 50 percent -- th e African-American population of Richmond. The Court in Croson demanded a more meaningful statistical comparison and much greater mathematical precision. It held that numerical figures used in a racial preference must bear a relationship to the pool of qualified minorities. Thus, in the Court's view, the thirty percent minority subcontracting requirement not narrowly tailored, because it was tied to the African-American population of Richmond, and as such, rested on the assumption that minorities will choose a particular trade "in lockstep proportion to their representation in the local population." 488 U.S. at 507. (52)

5. Duration and Periodic Review

Under Croson, affirmative action represents a "temporary" deviation from "the norm of equal treatment of all racial and ethnic groups." Croson, 488 U.S. at 510. A particular measure therefore should last only as long as it is needed. See Fullilove, 448 U.S. at 513 (Powell, J., concurring). Given this imperative, a racial or ethnic classification is more likely to pass the narrow tailoring test if it has a definite end-date, (53) or is subject to mean ingful periodic review that enables the government to ascertain the continued need for the measure. The Supreme Court has said that a set end-date is less important where a program does not establish specific numerical targets for minority participation. Johnson, 480 U.S. at 640. However, it remains important for such a program to undergo periodic review. See id. at 639-40.

Simply put, a racial or ethnic classification that was justified at the point of its adoption may no longer be required at some future point. If the classification is subject to reexamination from time to time, the government can react to changed circums tances by fine-tuning the classification, or discontinuing it if warranted. See Fullilove, 448 U.S. at 489 (plurality opinion); see also Metro Broadcasting, 497 U.S. at 594; Sheet Metal Workers, 478 U.S. at 478 (plurality opinion); id. at 4 87-88 (Powell, J., concurring).

6. Burden

Affirmative action necessarily imposes a degree of burden on persons who do not belong to the groups that are favored by a racial or ethnic classification. The Supreme Court has said, however, that some burdens are acceptable, even when visited upon indi viduals who are not personally responsible for the particular problem that the classification seeks to address. See Wygant, 476 U.S. at 280-81 (plurality opinion) ("As part of this Nation's dedication to eradicating racial discrimination, innocent person s may be called upon to bear some of the burden of the remedy."). This was implicitly reaffirmed in Croson and Adarand: in both cases, the Court "recognize[d] that any individual suffers an injury when he or she is disadvantaged by the government because of his or her race, whatever that race may be," (54)but declined to hold that the imposition of that burden pursuant to an affirmative action measure is automatically unconstitutional.

In some situations, however, the bu rden imposed by an affirmative action program may be too high. As a general principle, a racial or ethnic classification crosses that threshold when it "unsettle[s] . . . legitimate, firmly rooted expectation[s]," (55) or i mposes the "entire burden . . . on particular individuals." (56) Applying that principle in an employment case where seniority differences between minority and nonminority employees were involved, a plurality of the Court i n Wygant stated that race-based layoffs may impose a more substantial burden than race-based hiring and promotion goals, because "denial of a future employment opportunity is not as intrusive as loss of an existing job." Wygant, 476 U.S. at 282-83 ; see also id. at 294 (White, J., concurring). In a subsequent case, however, Justice Powell warned that "it is too simplistic to conclude that hiring [or other employment] goals withstand constitutional muster whereas layoffs do not . . . . The proper constitutional inquiry focuses on the effect, if any, and the diffuseness of the burden imposed on innocent nonminorities, not on the label applied to the particular employment plan at issue." Sheet Metal Workers, 478 U.S. at 488 n.3 (Powell, J., concurring).

In the contracting area, a racial or ethnic classification would upset settled expectations if it impaired an existing contract that had been awarded to a person who is not included in the classification. This apparently occurs rarely, if at all, in the federal government. A more salient inquiry therefore focuses on the scale of the exclusionary effect of a contracting program. For example, in Fullilove, Justice Powell thought it salient that the contracting requirement at issue in the case rese rved for minorities a very small amount of total funds for construction work in the nation (less than one percent), leaving nonminorities able to compete for the vast remainder. For Justice Powell, this rendered the effect of the program "limited and so widely dispersed that its use is consistent with fundamental fairness." Fullilove, 448 U.S. at 515. In some instances, conversely, the exclusionary effect of racial classifications in contracting may be considered too large. For example, the low er court in Croson held that Richmond's thirty percent minority subcontracting requirement imposed an impermissible burden because it placed nonminorities at a great "competitive disadvantage." J.A. Croson Co. v. City of Richmond, 822 F.2d 1355, 1361 (4th Cir. 1987). Similarly, an affirmative action program that effectively shut nonminority firms out of certain markets or particular industries might establish an impermissible burden. For example, the dissenters in Metro Broadcasting felt that the FCC's distress sale unduly burdened nonminorities because it "created a specialized market reserved exclusively for minority controlled applicants. There is no more rigid quota than a 100% set-aside . . . . For the would-be purchaser or person who seeks to compete for the station, that opportunity depends entirely upon race or ethnicity." 497 U.S. at 630 (O'Connor, J., dissenting). The dissenters also dismissed the majority's contention that the impact of distress sales on nonminoriti es was minuscule, given the small number of stations transferred through those means. The dissenters said that "[i]t is no response to a person denied admission at one school, or discharged from one job, solely on the basis of race, that other schools or employers do not discriminate." Id.

C. The Post-Croson Landscape at the State and Local Level

Croson has not resulted in the end of affirmative action at the state and local level. There is no doubt, however, that Croson, in tightening the constitutional parameters, has diminished the incidence of such programs, at least in contract ing and procurement. The post-Croson experience of governments that continue to operate affirmative action programs in that area is instructive. (57) Many governments reevaluated their MBE programs in light of Cr oson, and modified them to comport with the applicable standards. Typically, the centerpiece of a government's efforts has been a "disparity study," conducted by outside experts, to analyze patterns and practices in the local construction industry. The purpose of a disparity study is to determine whether there is evidence of discrimination against minorities in the local construction industry that would justify the use of remedial racial and ethnic classifications in contracting and procurement. So me studies also address the efficacy of race-neutral alternatives. In addition to obtaining a disparity study, some governments have held public hearings in which they have received evidence about the workings of the local construction industry.

Post-Croson affirmative action programs in contracting and procurement tend to employ flexible numerical goals and/or bidding preferences in which race or ethnicity is a "plus" factor in the allocation decision, rather than a hard set-aside of the sort at issue in Croson. It appears that many of the post-Croson contracting and procurement programs that rest on disparity studies have not been challenged in court. (58) At least one of the programs was sustained in litigation. (59) Another was struck down as inconsistent with the Croson standards. (60) Challenges to other programs were not resolved on summary judgment, and were remanded for f urther fact finding. (61) Contracting and procurement programs that were not changed after Croson have met with a mixed reception in the courts. (62)

III. Application of the Croson Standards at the Federal Level

In essence, Adarand federalizes Croson, with one important caveat: Congress may be entitled to some deference when it acts on the basis of race or ethnicity to remedy the effects of discrimination. The Court in Adarand hinted that a t least where a federal affirmative action program is congressionally mandated, the Croson standards might apply somewhat more loosely. The Court concluded that it need not resolve whether and to what extent the judiciary should pay special defere nce to Congress in this area. The Court did, however, cite the opinions of various Justices in Fullilove, Croson, and Metro Broadcasting concerning the significance of Congress' express constitutional power to enforce the antidiscrimination guarantees of the Thirteenth and Fourteenth Amendments -- under Section 2 of the former and Section 5 of the latter -- and the extent to which courts should defer to exercises of that authority that entail the use of racial and ethnic classifications to remedy discrimination. See 63 U.S.L.W. at 4531. Some of those opinions indicate that even under strict scrutiny, Congress does not have to make findings of discrimination with the same degree of precision as a state or local government, and that Congres s may be entitled to some latitude with respect to its selection of the means to the end of remedying discrimination. (63)

In Fullilove, Justice Powell's concurring opinion said that even under strict scrutiny, "[t]he degree of specificity required in the findings of discrimination and the breadth of discretion in the choice of remedies may vary with the nature and authority of a governmental body." Fullilove, 448 U.S. at 515 n.14 (Powell, J., concurring ). It was therefore of paramount importance to Justice Powell that the racial and ethnic classification in Fullilove was prescribed by Congress, which, Justice Powell admonished, "properly may -- and indeed must -- address directly the problems of discrimination in our society." Id. at 499. Justice Powell emphasized that Congress has "the unique constitutional power" to take such action under the enforcement clauses of the Thirteenth and Fourteenth Amendments. Id. at 500. See id. at 483 (plura lity opinion) ("[I]n no organ of government, state or federal, does there repose a more comprehensive remedial power than in the Congress, expressly charged by the Constitution with the competence and authority to enforce equal protection guarantees."). Justice Powell observed that when Congress uses those powers, it can paint with a broad brush, and can devise national remedies for the national problem of racial and ethnic discrimination. Id. at 502-03 (Powell, J., concurring). Furthermore, Justice Po well said that through repeated investigation of that problem, Congress has developed familiarity with the nature and effects of discrimination: "After Congress has legislated repeatedly in an area of national concern, its Members gain experience that ma y reduce the need for fresh hearings or prolonged debate when Congress again considers action in that area." Id. at 503. Because Congress need not redocument the fact and history of discrimination each time it contemplates adopting a new remedial measure , the findings that supported the Fullilove legislation were not restricted to the actual findings that Congress made when it enacted that measure. Rather, the record included "the information and expertise that Congress acquires in the considerat ion and enactment of earlier legislation." Id. A court reviewing a race-based remedial act of Congress therefore "properly may examine the total contemporary record of congressional action dealing with the problems of racial discrimination against [mino rities]." Id. Finally, Justice Powell gave similar deference to Congress when it came to applying the narrow tailoring test. He said that in deciding how best to combat discrimination in the country, the "Enforcement Clauses of the Thirteenth and Fourt eenth Amendments give Congress a . . . measure of discretion to choose a suitable remedy." Id. at 508.

Justice O'Connor's opinion in Croson is very much in the same vein. She too commented that Congress possesses "unique remedial powers . . . under § 5 of the Fourteenth Amendment." Croson, 488 U.S. at 488 (plurality opinion) (citing Ful lilove, 448 U.S. at 483 (plurality opinion)). By contrast, state and local governments have "no specific constitutional mandate to enforce the dictates of the Fourteenth Amendment," but rather are subject to its "explicit constraints." Id. at 490 (p lurality opinion). Therefore, in Justice O'Connor's view, state and local governments "must identify discrimination, public or private, with some specificity before they may use race-conscious relief." Id. at 504. Congress, on the other hand, can make, and "has made national findings that there has been societal discrimination in a host of fields." Id. It may therefore "identify and redress the effects of society-wide discrimination" through the use of racial and ethnic classifications that would be impermissible if adopted by a state or local government. Id. at 490 (plurality opinion). (64) Justice O'Connor cited her Croson opinion and reiterated these general points about the powers of Congress in her Met ro Broadcasting dissent. See 497 U.S. at 605 (O'Connor, J., dissenting) ("Congress has considerable latitude, presenting special concerns for judicial review, when it exercises its unique remedial powers . . . under § 5 of the Fourteenth Amendment.") (internal quotations omitted).

It would be imprudent, however, to read too much into Justice Powell's opinion in Fullilove and Justice O'Connor's opinion in Croson. They do not, for example, support the proposition that Congress may simply assert that because there has b een general societal discrimination in this country, legislative classifications based on race or ethnicity are a necessary remedy. The more probable construction of those opinions is that Congress must have some particularized evidence about the existen ce and effects of discrimination in the sectors and industries for which it prescribes racial or ethnic classifications. For example, Congress established the Fullilove racial and ethnic classification to remedy what the Court saw as the well-docu mented effects of discrimination in one industry -- construction -- that had hindered the ability of minorities to gain access to public contracting opportunities. See Fullilove, 448 U.S. at 505-06 (Powell, J., concurring); see also id. at 473 (p lurality opinion).

Based on this reading of Croson and Fullilove, the endorsement in Adarand of strict scrutiny of federal affirmative action programs does not mean that Congress must find discrimination in every jurisdiction or industry affected by suc h a measure (although it is unclear whether, as a matter of narrow tailoring, the scope of a classification should be narrowed to exclude regions and trades that have not been affected by the discrimination that is to be remedied.). State and local gover nments must identify discrimination with some precision within their jurisdictions; Congress' jurisdiction is the nation as a whole. But after Adarand, Congress is subject to the Croson "strong basis in evidence" standard. Under that stand ard, the general history of racial discrimination in the nation would not be a sufficient predicate for a remedial racial or ethnic classification. In addition, evidence of discrimination in one sector or industry is not always probative of discriminatio n in other sectors and industries. For example, a history of lending discrimination against minorities arguably cannot serve as a catch-all justification for racial and ethnic classifications benefitting minority-owned firms through the entire economy; a pplication of the narrow tailoring test would suggest that if lending discrimination is the problem being addressed, then the government should tackle it directly. (65)

Furthermore, under the new standard, Congress pro bably does not have to hold a hearing or draft a report each time it adopts a remedial racial or ethnic classification. But where such a classification rests on a previous law or series of laws, those earlier measures must be supported by sufficient evi dence of the effects of discrimination. And if the findings in the older laws are stale, Congress or the pertinent agency may have to demonstrate the continued relevance of those findings; this would satisfy the element of the narrow tailoring test that looks to the duration of classifications and whether they are subject to reevaluation. Where the record is sparse, Congress or the relevant agency may have to develop it. That endeavor may involve the commissioning of disparity studies of the type that state and local governments around the country undertook after Croson to demonstrate that remedial racial and ethnic classifications in public contracting are warranted. Together, the myriad state and local studies may provide an important source of evidence supporting the use by the federal government of national remedial measures in certain sectors of the economy.

Whatever deference a court might accord to federal remedial legislation after Adarand, it is undecided whether the same degree of deference would be accorded to nonremedial legislation. In Metro Broadcasting, the majority gave substantial d eference to congressional judgments regarding the need for diversity in broadcasting and the linkage between the race of a broadcaster and programming output. Metro Broadcasting, 497 U.S. at 566, 572-73, 591 n.43. The dissenters did not do so, p recisely because the classifications were nonremedial and hence, in their view, did not implicate Congress' powers under the Enforcement Clauses of the Thirteenth and Fourteenth Amendments. Id. at 605, 628-29 (O'Connor, J., dissenting).

Finally, many existing federal affirmative action programs are not specifically mandated by Congress. Courts are unlikely to accord federal agencies acting without a congressional mandate the same degree of deference accorded judgments made by Congress i tself. Agencies do not have the "institutional competence" and explicit "constitutional authority" that Congress possesses. Adarand, 63 U.S.L.W. at 4538 (Stevens, J., dissenting). (66) Although some existing agency programs were not expressly mandated in the first instance in legislation, they may nonetheless be viewed by a court as having been mandated by Congress through subsequent congressional action. For example, in Metro Broadcasting, the programs at issue were established by the FCC on its own; Congress' role was limited to FCC oversight hearings and the passage of an appropriations riders that precluded the FCC from using any funds to reconsider or cancel its programs. 497 U.S. at 572-79. The majo rity concluded that this record converted the FCC programs into measures that had been "specifically approved -- indeed, mandated by Congress." Id. at 563.

Under strict scrutiny, it is uncertain what level of congressional involvement is necessary before a court will review an agency's program with deference. What may be required is evidence that Congress plainly has brought its own judgment to bear on the matter. Cf. Adarand, 63 U.S.L.W. at 4537 (Stevens, J., dissenting) ("An additional reason for giving greater deference to the National Legislature than to a local law-making body is that federal affirmative-action programs represent the will of our entire Nation's elected representatives . . . .") (emphasis added); id. at 4538 (Stevens, J., dissenting) ("Congressional deliberations about a matter as important as affirmative action should be accorded far greater deference than those o f a State or municipality.") (emphasis added).

IV. Conclusion

Adarand makes it necessary to evaluate federal programs that use race or ethnicity as a basis for decisionmaking to determine if they comport with the strict scrutiny standard. No affirmative action program should be suspended prior to such an eva luation. The information gathered by many agencies in connection with the President's recent review of federal affirmative action programs should prove helpful in this regard. In addition, appended to this memo is a nonexhaustive checklist of questions that provides initial guidance as to what should be considered in that review process. Because the questions are just a guide, no single answer or combination of answers is necessarily dispositive as to the validity of any given program.

Appendix: Questions to Guide Review
of Affirmative Action Programs

I. Authority

Is the use of racial or ethnic criteria as a basis for decisionmaking mandated by legislation? If not mandated, is it expressly authorized by legislation? If there is no express authorization, has there been any indication of congressional approval of an agency's action in the form of appropriations riders or oversight hearings? These questions are important, because Congress may be entitled to some measure of deference when it decides that racial and ethnic classifications are necessary.

If there is no explicit legislative mandate, authorization, or approval, is the program premised on an agency rule or regulation that implements a statute that, on its face, is race-neutral? For example, some statutes require agencies to give preference s to "disadvantaged" individuals, but do not establish a presumption that members of racial groups are disadvantaged. Such a statute is race-neutral. Other statutes, like those at issue in Adarand, require agencies to give preferences to "disadva ntaged" individuals, but establish a rebuttable presumption that members of racial groups are disadvantaged. Such a statute is race-conscious, because it authorizes agencies to use racial criteria in decisionmaking.

II. Purpose

What is the objective of the program? Is it intended to remedy discrimination, to foster racial diversity in a particular sector or industry, or to achieve some other purpose? Is it possible to discern the purpose from the face the relevant statute or legislation? If not, does the record underlying the relevant legislation or regulation shed any light on the purpose of the program?

A. Factual Predicate: Remedial Programs

If the program is intended to serve remedial objectives, what is the underlying factual predicate of discrimination? Is the program justified solely by reference to general societal discrimination, general assertions of discrimination in a particular se ctor or industry, or a statistical underrepresentation of minorities in a sector or industry? Without more, these are impermissible bases for affirmative action. If the discrimination to be remedied is more particularized, then the program may satisfy < b>Adarand. In assessing the nature of the factual predicate of discrimination, the following factors should be taken into account:

1. Source. Where can the evidence be found? Is it contained in findings set forth in a relevant statute or legislative history (committee reports and hearings)? Is evidence contained in findings that an agency has made on its own in connection with a rulemaking process or in the promulgation of guidelines? Do the findings expressly or implicitly rest on findings made in connection with a previous, related program (or series of programs)?

2. Type. What is the nature of the evidence? Is it statistical or documentary? Are the statistics based on minority underrepresentation in a particular sector or industry compared to the general minority population? Or are the statistics more sophisticated and focused? For example, do they attempt to identify the number of qualified minorities in the sector or industry or seek to explain what that number would look like "but for" the exclusionary effects of discrimination? Does the evidence seek to explain the secondary effects of discrimination -- for example, how the inability of minorities to break into certain industries due to historic practices of exclusion has hindered their ability to acquire the requisite capital and financing? Si milarly, where health and education programs are at issue, is there evidence on how discrimination has hampered minority opportunity in those fields, or is the evidence simply based on generalized claims of societal discrimination? In addition to any sta tistical and documentary evidence, is there testimonial or anecdotal evidence of discrimination in the record underlying the program -- for example, accounts of the experiences of minorities and nonminorities in a particular field or industry?

3. Scope. Are the findings purported to be national in character and dimension? Or do they reflect evidence of discrimination in certain regions or geographical areas?

4. "Authorship". If Congress or an agency relied on reports and testimony of others in making findings, who is the "author" of that information? The Census Bureau? The General Accounting Office? Business and trade associations? Academic exp erts? Economists? (There is no necessary hierarchy in assessing authorship, but the identity of the author may affect the credibility of the findings.)

5. Timing. Since the adoption of the program, have additional findings of discrimination been assembled by Congress or the agency that could serve to justify the need for the program when it was adopted? If not, can such evidence be readily ass embled now? These questions go to whether "post-enactment" evidence can be marshaled to support the conclusion that remedial action was warranted when the program was first adopted.

B. Factual Predicate: Nonremedial Programs

Adarand does not directly address whether and to what extent nonremedial objectives for affirmative action may constitute a compelling governmental interest. At a minimum, to the extent that an agency administers a nonremedial program intended to promote diversity, the factual predicate must show that greater diversity would foster some larger societal goal beyond diversity for diversity's sake. The level and precision of empirical evidence supporting that nexus may vary, depending on the nature and purpose of a nonremedial program. For a nonremedial program, the source, type, scope, authorship, and timing of underlying findings should be assessed, just as for remedial programs.

III. Narrow Tailoring

A. Race-Neutral Alternatives

Did Congress or the agency consider race-neutral means to achieve the ends of the program at the time it was adopted? Race-neutral alternatives might include preferences based on wealth, income, education, family, geography. In the commercial setting, another such alternative is a preference for new, emerging businesses. Were any of these alternatives actually tried and exhausted? What was the nature and extent of the deliberation over any race-neutral alternatives -- for example, congressional debat e? agency rulemaking? Was there a judgment that race-neutral alternatives would not be as efficacious as race-conscious measures? Did Congress or the agency rely on previous consideration and rejection of race-neutral alternatives in connection with a p rior, related race-conscious measure (or series of measures)?

B. Continued Need

How long has the program been in existence? Even if there was a compelling justification at the time of adoption, that may not be the case today. Thus, an agency must determine whether there is a continued need for the program. In that regard, does th e program have an end date? Has the end date been moved back? Is the program subject to periodic oversight? What is the nature of that oversight -- does Congress play a role through hearings/reports, or does the agency conduct the review or oversight on its own? Has the program ever been adjusted or modified in light of a periodic review? What were the results of the most recent review and oversight conducted by either Congress or the agency? Is there evidence of what might result if the racial cl assification were discontinued? For example, is there evidence of the current level of minority participation in government contracting where racial criteria are not used (which may speak to whether discrimination can be remedied without a preference)?

C. Pool of Beneficiaries

Are the benefits of the program spread relatively equally among minority individuals or businesses? Is there information on whether the same individuals or businesses tend to reap most of the benefits, and if so, whether those beneficiaries have overcom e discrimination? If the program is intended to remedy discrimination against minorities, does it include among its beneficiaries subgroups that may not have been discriminated against? Is there a procedure for tailoring the pool of beneficiaries to exc lude such subgroups? Is there a mechanism for evaluating whether the program is needed for segments within a larger industry that have been the locus of discrimination?

D. Manner in Which Race is Used

Does the program establish fixed numerical set-asides? Is race an explicit requirement of eligibility for the program? If there is no such facial requirement, does the program operate that way in practice? Or is race just one of several factors -- a " plus" -- used in decisionmaking? Could the objectives of a program that uses race as a requirement for eligibility be achieved through a more flexible use of race?

E. Burden

What is the nature of the burden imposed on persons who are not included in the racial or ethnic classification that the program establishes? Does the program displace those persons from existing positions/contracts? Does it upset any settled expectati ons that they have? Even if that is not the case, the burden may be impermissible where the exclusionary impact is too great. What is the exclusionary impact in terms of size and dimension? What is the dollar value of the contracts/grants/positions in question? Does the exclusionary impact of the program fall upon a particular group or class of individuals or sectors, or is it more diffuse? What is the extent of other opportunities outside the program? Are persons who are not eligible for the prefer ence put at a significant competitive disadvantage as a result of the program?